This is an automated email from the ASF dual-hosted git repository.
linkinstar pushed a commit to branch dev
in repository https://gitbox.apache.org/repos/asf/answer.git
The following commit(s) were added to refs/heads/dev by this push:
new a15dd415 refactor(internal): compile regex once while clearing text
a15dd415 is described below
commit a15dd41550c282a4de44259c69a57dd52eeb0e9c
Author: ferhat elmas <[email protected]>
AuthorDate: Sat Nov 22 12:43:55 2025 +0100
refactor(internal): compile regex once while clearing text
Regex and Replacer can be reused.
No need to recreate for each invocation.
Signed-off-by: ferhat elmas <[email protected]>
---
pkg/htmltext/htmltext.go | 48 ++++++++++++++++++++++--------------------------
1 file changed, 22 insertions(+), 26 deletions(-)
diff --git a/pkg/htmltext/htmltext.go b/pkg/htmltext/htmltext.go
index 49049109..0b84cda4 100644
--- a/pkg/htmltext/htmltext.go
+++ b/pkg/htmltext/htmltext.go
@@ -34,38 +34,34 @@ import (
"github.com/mozillazg/go-pinyin"
)
+var (
+ reCode = regexp.MustCompile(`(?ism)<(pre)>.*<\/pre>`)
+ reCodeReplace = "{code...}"
+ reLink = regexp.MustCompile(`(?ism)<a.*?[^<]>(.*)?<\/a>`)
+ reLinkReplace = " [$1] "
+ reSpace = regexp.MustCompile(` +`)
+ reSpaceReplace = " "
+
+ spaceReplacer = strings.NewReplacer(
+ "\n", " ",
+ "\r", " ",
+ "\t", " ",
+ )
+)
+
// ClearText clear HTML, get the clear text
-func ClearText(html string) (text string) {
- if len(html) == 0 {
- text = html
- return
+func ClearText(html string) string {
+ if html == "" {
+ return html
}
- var (
- re *regexp.Regexp
- codeReg = `(?ism)<(pre)>.*<\/pre>`
- codeRepl = "{code...}"
- linkReg = `(?ism)<a.*?[^<]>(.*)?<\/a>`
- linkRepl = " [$1] "
- spaceReg = ` +`
- spaceRepl = " "
- )
- re = regexp.MustCompile(codeReg)
- html = re.ReplaceAllString(html, codeRepl)
+ html = reCode.ReplaceAllString(html, reCodeReplace)
+ html = reLink.ReplaceAllString(html, reLinkReplace)
- re = regexp.MustCompile(linkReg)
- html = re.ReplaceAllString(html, linkRepl)
-
- text = strings.NewReplacer(
- "\n", " ",
- "\r", " ",
- "\t", " ",
- ).Replace(strip.StripTags(html))
+ text := spaceReplacer.Replace(strip.StripTags(html))
// replace multiple spaces to one space
- re = regexp.MustCompile(spaceReg)
- text = strings.TrimSpace(re.ReplaceAllString(text, spaceRepl))
- return
+ return strings.TrimSpace(reSpace.ReplaceAllString(text, reSpaceReplace))
}
func UrlTitle(title string) (text string) {
